Research map / Architectures and efficiency / Neural architecture search
NAS-DIP: Learning Deep Image Prior with Neural Architecture Search
Yun-Chun Chen, Chen Gao, Esther Robb, Jia-Bin Huang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add searched cross-scale residual connections to searched upsampling (ablation) | NAS-DIP · Ablation: Set14 SR 4x | PSNR (dB): 27.54 -> 27.84 | Improved |
| add searched cross-scale residual connections to searched upsampling (ablation) | NAS-DIP · Ablation: inpainting | PSNR (dB): 34.01 -> 34.72 | Improved |
| add searched cross-scale residual connections to searched upsampling (ablation) | NAS-DIP · Ablation: denoising | PSNR (dB): 31.08 -> 31.42 | Improved |
| add searched upsampling cell to searched connections (ablation) | NAS-DIP · Ablation: Set14 SR 4x | PSNR (dB): 27.32 -> 27.84 | Improved |
| add searched upsampling cell to searched connections (ablation) | NAS-DIP · Ablation: Set14 SR 8x | PSNR (dB): 24.29 -> 24.59 | Improved |
| add searched upsampling cell to searched connections (ablation) | NAS-DIP · Ablation: inpainting | PSNR (dB): 34.59 -> 34.72 | Improved |
| add searched upsampling cell to searched connections (ablation) | NAS-DIP · Ablation: denoising | PSNR (dB): 31.16 -> 31.42 | Improved |
| use searched architecture instead of hand-crafted U-Net (DIP) | DIP · Set14 super-resolution 4x | PSNR (dB): 27.00 -> 27.84 | Improved |
| use searched architecture instead of hand-crafted U-Net (DIP) | DIP · Image denoising (BM3D dataset) | PSNR (dB): 30.43 -> 31.42 | Improved |
| use searched architecture instead of Deep Decoder | Deep Decoder · Deep Decoder protocol: inpainting | PSNR (dB): 31.9 -> 33.1 | Improved |
| use searched model in CycleGAN instead of standard U-Net | CycleGAN · Summer-to-Winter translation | FID: 78.62 -> 76.22 | Improved |
| use searched model in CycleGAN instead of standard U-Net | CycleGAN · Winter-to-Summer translation | FID: 73.91 -> 71.98 | Improved |
Closest papers by meaning
- ISNAS-DIP: Image-Specific Neural Architecture Search for Deep Image Prior (CVPR 2022 · similarity 0.94)
- Memory-Efficient Hierarchical Neural Architecture Search for Image Restoration (IJCV 2021 · similarity 0.90)
- CLEARER: Multi-Scale Neural Architecture Search for Image Restoration (NeurIPS 2020 · similarity 0.89)
- A Restoration Network as an Implicit Prior (ICLR 2024 · similarity 0.88)
- UGoDIT: Unsupervised Group Deep Image Prior Via Transferable Weights (NeurIPS 2025 · similarity 0.87)
- L-SWAG: Layer-Sample Wise Activation with Gradients Information for Zero-Shot NAS on Vision Transformers (CVPR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
